How Do External Factors Influence Passenger Rail Demand?
A Meaney, M Shepherd, A Horncastle, Oxera Consulting Ltd, UK; J Cartmell, Department for Transport, UK; W Wingate, Arup, UK
How do factors such as income, demographics, and costs of other modes of transport affect passenger rail demand? These are studied using a new panel dataset, innovative market segmentation and cutting edge dynamic panel data econometric techniques.
At ETC in 2009, Arup, Oxera and the Department for Transport presented new work on how the passenger rail market should be segmented for the purposes of developing a forecasting framework for GB. This paper presents the new forecasting framework, which has now been developed following further analysis of the dataset using the latest dynamic panel data econometric techniques on a large dataset of rail demand (over 20,000 origin-destination pairs, covering six ticket types, over 18 years). As far as the study team is aware, this is the first time that dynamic panel data econometrics has been used for such an analysis, with several innovations in transport analysis contained within the paper.
The paper focuses on the econometric analyses that have been carried out on the segmented dataset, the findings of those analyses, the forecasting framework that has been developed, and some of the policy implications emerging from the study.
The study has permitted the simultaneous estimation of elasticities of rail demand to different measures of income, fares, demographics (population and employment), car costs and journey times, rail journey times, rail punctuality and reliability, and rail service quality (using a new service quality index developed for the study).
Following on from the market segmentation work presented at ETC in 2009, the study team applied the latest dynamic panel data econometric techniques to the data. The adopted process enabled the team to assess a number of hypotheses, including the existence of lagged effects on each variable; whether elasticities change for different levels of the explanatory variables; whether elasticities changed by journey length (distance); whether elasticities changed over time, and whether there is any evidence of market saturation (ie, whether the link between rail demand and GDP is decreasing over time). All of these potential effects were investigated simultaneously for the first time in this study.
This process was adopted for each of 28 ?ticket type market segments??broadly, journeys between and within three geographies, using three ticket types (with some additional coverage). This paper presents elasticities from each of the market segments, and also reviews the extent to which the hypotheses above are rejected in the dataset. Furthermore, the paper highlights which measures of the explanatory variables can explain variations in rail demand. This has particular implications for income elasticities.
The paper will also present key elements of the forecasting framework, including the appropriate treatment of dynamic (lagged) effects, of variable elasticities, and market saturation effects for spreadsheet-based demand forecasting.
The paper will conclude with some interesting policy implications arising from the analysis, including new evidence on fare, rail journey time, and punctuality elasticities; on links between the characteristics of car travel and rail demand; and on market saturation.
Association for European Transport